I want to pick a very small fight with Paul Krugman. And I want to get my head clearer on something. (It's still not clear, so read at your own risk.) So I'm writing this post.
"The problem, at least in part, is that the indirectness of Bernanke’s
language, the way an inflation target is implicit rather than explicit,
feeds the confusion. I understand what he’s doing: a lot of people
aren’t ready to face the realities here, so blurriness has its uses (and
so do other targets, like nominal GDP, that are ultimately mainly about
inflation but don’t make that point explicitly). But there are costs to
this vagueness, and we’re seeing some of them already." (my bold)
Agreed that the Fedspeak is not as clear as it should be. But I don't think that an NGDP target really is ultimately mainly about inflation. A higher NGDP target is like a portfolio of inflation and real growth, and when you are uncertain a portfolio of two assets is usually better than a single asset. Eggs and baskets stuff.
There are four things we don't know:
1. We don't know the slope of the Short Run Phillips Curve. And we don't know what people believe about the slope of the Short Run Phillips Curve. If the SRPC is (believed to be) steep, an increase in (expected future) NGDP will mean a large increase in (expected) inflation and a small increase in (expected) real growth. If the SRPC is (believed to be) flat, an increase in (expected future) NGDP will mean a small increase in (expected) inflation and a large increase in (expected) real growth.
2. For a given nominal interest rate, and increase in expected inflation will cause a decrease in real interest rates and an increase in Aggregate Demand. For a given real interest rate, an increase in expected real growth will cause an increase in Aggregate Demand (because people and firms want to consume more and invest more if they think future real incomes will be higher). But we don't know the relative strengths of these two effects.
3. An increase in expected inflation will cause an upward shift in the Short Run Phillips Curve. In the Calvo model a 1% increase in expected inflation will cause an immediate 1% upward shift in the SRPC. In a different model, with more inflation inertia, it will take longer for a 1% increase in expected inflation to cause a 1% upward shift in the SRPC. And we don't know which model is right, and so we don't know how quickly and how much the SRPC will shift up if expected inflation increases by 1%. (Though we think we know it will eventually shift up by exactly 1%.)
4.Short run "supply shocks" (perhaps better described as "price shocks") can cause the Short Run Phillips Curve to shift independently of any shift they cause in the Long Run Phillips Curve. We do not have a good theory of how short run supply shocks shift the SRPC. And we do not know what short run supply shocks there will be in the near future. So we do not know how big an increase in inflation would be consistent with getting the economy to the Long Run Phillips Curve.
If we did know all those four things, It wouldn't matter which language the Fed spoke. We would know how much inflation would be associated with a higher level of NGDP, and vice versa. We could translate inflation Fedspeak into NGDP Fedspeak, and translate NGDP Fedspeak into inflation Fedspeak. It wouldn't be correct to say that an NGDP target is "ultimately mainly about inflation". Nor would it be correct to say that an inflation target is "ultimately mainly about NGDP". It would be correct to say they are different ways of saying the same thing.
But we don't know those four things. And even if we economists did know all those four things, regular people might not. And the main point of Fedspeak is to communicate with regular people. So we cannot translate back and forth between an inflation target and an NGDP target. The Fed's language matters.
Which language is best?
I wish I could build a neat little model with parameter uncertainty to show which is best. But I can't. I can only sketch a heuristic argument.
An inflation target to escape the Zero Lower Bound seems like putting all your eggs in one basket. Expected inflation goes up, expected real interest rates go down, Aggregate Demand goes up, actual inflation goes up, and validates the increase in expected inflation. And real income goes up, maybe by just the right amount, or maybe too little, or maybe too much. It could be far too little, or far too much, if we are wrong about the slope and position of the Short Run Phillips Curve, or about the strength of the real interest rate effect.
An NGDP target is more like putting half your eggs in the inflation basket and half your eggs in the real growth basket. One basket should work, even if the other fails. The two baskets are even negatively correlated, via the uncertain slope of the Short Run Phillips Curve. It should be more likely to work, and work by roughly the right amount.
That's the best I can do, for now.
Worthwhile Canadian Initiative 
This actually brings up a question I’ve always had. Why is it that in the history of economic thought, macro models have wages/prices set based on ONLY expected inflation, and not rGDP as well. In Retrospect, there is no logical reason to believe rGDP doesn’t get factored in by firms and households. Why did economists decided to do expected versus actual INFLATION, and not expected versus actual NGDP?
You would think that somebody in 100 years would have attempted this.
JoeMac, I’m no expert on the history of economic thought, but there’s a huge exception to your claim regarding macro models. Keynes’s General Theory is presented in terms of spending expressed in money and/or wage-units.
Great post.
I suppose it was natural I was troubled by the same Krugman post, thought much along the same lines, and remembers your earlier post using the diversification argument.
I think Krugman would say that nominal GDP is about inflation because the central bank is already committed to closing the output gap. And so, any appropriate increase in future real real income and its effect on current spending is already included. Any added impact of expected higher nominal GDP on current expenditure must either be due to real output rising above potential or else higher inflation. A future boom.
As level targeters, most Market Monetarists would say that if inflation fell below target, it should be returned to the previous growth path, which require a higher inflation rate. But, of course, this can be described as being about the inflation.
However, one puzzle is that higher expected inflation raises aggregate demand due to a lower real interest rate, but it reduces aggregate supply–as you described in the post. A committmeent to higher inflation may cause more spending, but will tend to reduce how much of that spending generates real output and employment. A committment to more rapid spending growth makes increased spending now now sensible, without the central bank telling everyone to raise prices faster, which tends to depress output growth.
Higher inflation won’t cause us to give up on our effortt to raise spending on output, but we aren’t going to do what it takes to make sure inflation is higher, which would presumably might involve accelerating spending growth.
Nick
Good post. Completely agree that an NGDP target is not like an inflation target, and is actually superior. Krugman has clearly accepted Wicksell and the Fisher relation, which is why he talks about inflation and real rates. He only needs to connect the final bit about the Wicksellian natural rate being highly correlated with the growth rate of the economy and the superiority of an NGDP target will show itself (irrespective of debates about central bank effectiveness in necessarily hitting the target).
I guess Krugman believes that there is ‘a’ natural growth rate of the economy which is 3%. And the Wicksellian rate can be whatever it is. I tend to agree with you that there’s a very strong correlation. A 5% NGDP target makes sense because it is ‘sufficiently’ above 0% and yet sufficiently ‘stable’. One doesn’t need to believe any specific potential growth rate of the economy. The 3% + 2% breakdown is irrelevant.
What robust empirical evidence supports the notion that there IS a short run Phillips Curve? And, in the absence of such evidence, why would one continue to build models and/or discuss the performance of the aggregate economy AS IF a short run Phillips Curve were crucial to explaining variations in its key variables?
JoeMac: IIRC (I may not) the individual firm in the Calvo model sets its price based on both expected inflation and on expected future real demand for its product. But when you solve out for the equilibrium, only expected inflation and current real demand remain in the Phillips Curve. In a model with inflation inertia, I think future real demand would appear in the equation too. But those models are too complicated for analytical work.
Bill: Thanks! Yes. I think you are right on that point, and I left it out. The currency reform is the classic example where the SRPC is vertical, because the government announces a focal point for prices, and everyone multiplies or divides their price by 10 on the same day, because they expect everyone else to do the same.
Ritwik: Thanks!
Hi. Does quality uncertainty matter as well? Most people would be surprised to find that real growth in economic sectors like fiber optic data transmission has been low over the past decade (even as they’re consuming 50GB a month in video). Would we be at risk of overemphasizing both inflation, and arbitrary real growth over quality? I guess this is another way of saying that people may be suspicious of Fed speak focusing on raising housing prices over other things (e.g. investments in low MPG transportation).
Excellent post. Not surprisingly I had the same general reaction when I read Krugman’s comment, but wouldn’t have been able to express my objections as well as you did.
Why does Nick always say “Philips Curve”; wouldn’t it be more accurate here to say “Aggregate Supply Curve”?
Bill’s comment was good, but the last part confused me. Doesn’t price stickiness hinder the upward shift of SRAS due to expected inflation? Once the negative demand shock becomes expected, SRAS shifts down, but not all the way due to price stickiness. Then restoring the NGDP trendline would simply reverse that demand shock, and also SRAS would shift up correspondingly, not all the way. And I couldn’t make any sense out of the last sentence.
I left a comment on Clive Crook’s article at the Atlantic (http://www.theatlantic.com/business/archive/2012/09/where-the-fed-stands-on-monetary-policy/262426/#comment-654425776), where he endorsed NGDP growth targeting but not level targeting, due to Krugmanite “promising to be irresponsible” concerns. In it, I brought up the point that boosting future NGDP at the point where the AS has returned to long run equilibrium and the ZLB no longer binds, would also raise NGDP levels up to that point through asset price effects. But even if that “threatened” future boost to NGDP was all inflation, if it succeeded in boosting NGDP up to the trendline, pulling current NGDP up via real growth and unemployment elimination, and we followed that restored level path up to the future date, by the time that “threatened” period actually arrived there would be no inflation, as the counterfactual deflationary AS shift would have been averted.